基于MapReduce面向列的数据库存储方案研究
Research on Column-Oriented Database Storage Scheme Based on MapReduce
-
摘要: 将传统的并行DBMS技术应用到Hadoop框架,在此基础上将面向列的数据库存储技术引入Hadoop的复制和调度机制,进而获得高效的编程模型与编程API.实验表明,提出的方案主要应用在map phase阶段,同时能够使得MapReduce的性能提高两倍.Abstract: To improve the performance of MapReduce,the paper introduces the parallel DBMS into the Hadoop framework and introduces the column-oriented storage technology into the copy and scheduling of Hadoop,to get efficient coding model.Experimental results show the scheme presented in the paper can improve the performance of MapReduce.